International Journal of Business Intelligence and Systems Engineering (5 papers in press)
GENETIC ALGORTHIM BASED HYBRID APPROACH FOR OPTIMAL INSTANCE SELECTION OF MINIMIZING MAKESPAN IN PERMUTATION FLOWSHOP SCHEDULING
by Balasundaram Rathinam, Sathiya Devi S
Abstract: Recently, the instance selection is getting more attention for the researchers to achieve enhanced performance of the algorithm. A typical flowshop dataset can be represented in the form of a number of instances. The Instances that are recorded during production process may not be a good example to learn useful knowledge. Therefore, the selection of high quality instances can be considered as a search problem and solved by evolutionary algorithms. In this work, Genetic Algorithm (GA) is proposed to select sub-set of best instances. The selected instances (low level data) are represented in the form of IF-Then else rules (high level knowledge) using Decision Tree (DT) algorithm. Since, the DT is heuristics the seed solution generated from DT may not yield a good solution. Hence, seed solution from DT is given to input of Scatter Search (SS) algorithm for few iterations which acts as a local search to find the best value of the selected instances. The GA is used to select best instances in order to have a less tree size (number of rules are minimum) with good solution accuracy for minimizing makespan criterion in permutation flowshop scheduling. The computational experiments are performed with standard problems and compared against various existing literatures. Statistical tests of significance are performed to verify the development in solution.
Keywords: Instance selection; Genetic algorithm; Decision Tree algorithm; Makespan;.
Benchmarking the operating efficiency of US regional banks
by D.K. Malhotra, Rashmi Malhotra, Ruben A. Mendoza
Abstract: Banks, as private entities with a public purpose, strike a balance by achieving high profitability levels while managing the risks involved. Thus, understanding the factors that drive an efficiently-operating bank is an important issue. This study illustrates the use of data envelopment analysis (DEA) methodology to benchmark the operating efficiency of 34 US regional banks during the period 2009 to 2013. The study finds that only three banks out of 34 banks were 100% efficient relative to others throughout the sample period of 2009 to 2013. In addition, we also use the DEA's slack analysis to understand the factors responsible for the poor performance of a bank. Finally, we also investigate the factors contributing to the performance of the banking industry by covering a time period spanning the start of the economic crisis and the consequent passing of new laws to regulate the financial services industry.
Keywords: regional banks; data envelopment analysis; DEA; benchmarking; operating efficiency.
Operational intelligence through performance trends: an oracle prototype
by Rajeev Kaula
Abstract: Improving business process performance through operational intelligence is essential to enhance an organisation's ability to achieve business objectives and competitive advantage. This paper proposes an extension on insights provided by traditional business intelligence analytics through star schema that goes beyond snapshots of business process performance to outlining details on factors influencing business process operations over a period of time. Such extended insights, referred as performance trends, enable a richer assessment of business process performance. Performance trends essentially allow a business to determine whether the direction of business process performance is going up, down, or staying flat with respect to some success measure over a period of time. It is expressed as an analytic business rule that can be applied on the business process through analytic triggers. The paper illustrates the concepts through a prototype that is adapted from the oracle's e-business suite lead to forecast business process. The prototype is implemented in oracle's PL/SQL language.
Keywords: business intelligence; operational intelligence; business process; business rules; oracle; PL/SQL.
A comparative analysis of classifiers in cancer prediction using multiple data mining techniques
by Seyed Mohammad Jafar Jalali, Sérgio Moro, Mohammad Reza Mahmoudi, Keramat Allah Ghaffary, Mohsen Maleki, Aref Alidoostan
Abstract: In recent years, application of data mining methods in health industry has received increased attention from both health professionals and scholars. This paper presents a data mining framework for detecting breast cancer based on real data from one of the Iran hospitals by applying association rules and the most commonly used classifiers. The former were adopted for reducing the size of datasets, while the latter were chosen for cancer prediction. A k-fold cross-validation procedure was included for evaluating the performance of the proposed classifiers. Among the six classifiers used in this paper, support vector machine achieved the best results, with an accuracy of 93%. It is worth mentioning that the approach proposed can be applied for detecting other diseases as well.
Keywords: cancer prediction; data mining; classifiers; association rules.
Evaluating bank solvency with support vector machines
by D.K. Malhotra, Robert L. Nydick, Kunal Malhotra
Abstract: Banks as financial intermediaries play a very useful role in economic growth by facilitating the flow of funds to various sectors of the economy. Deterioration in a bank's performance and potential failure of the bank may lead to loss of confidence in the financial system that can result in loss of household savings and non-availability of funds to the business sector for economic expansion and growth. Banking regulators around the world are always looking for ways to identify sooner the banks that can be at risk of failure so that corrective action can be taken with minimal disruption to the economy. This study illustrates the use of support vector machines, an artificial intelligence technique, to predict the pending insolvency of a bank so that regulators can take appropriate steps to prevent a 'domino effect'. The study also compares the performance of support vector machines to multiple discriminant analysis in identifying 'unsafe' banks. To alleviate the problem of bias in the training set and to examine the robustness of support vector machine classifiers in identifying unsafe banks, we cross-validate our results through seven different samples of the data.
Keywords: bank failure; AI; support vector machines; SVM.